Table 3

Numerical comparison of analysis methods applied to the real data

Comb. (kest)
 Comb. (kfix = 0.5)
 Comb. (kfix = 0.25)
 Unaffected
 Single Channel
 SC (damaged)
 Inclusive
Combined(kest)
 -
 0.88
 0.89
 0.83
 0.70
 0.62
 0.57
Combined(kfix = 0.5)
 904
 -
 0.98
 0.80
 0.78
 0.71
 0.56
Combined(kfix = 0.25)
 842
 922
 -
 0.88
 0.74
 0.66
 0.57
Unaffected Data
 669
 723
 788
 -
 0.58
 0.48
 0.53
Single Channel
 750
 725
 686
 575
 -
 0.84
 0.56
SC (damaged arrays)
 710
 656
 612
 492
 788
 -
 0.53
Inclusive
 620
 629
 631
 591
 660
 605
 -

For every pair of analyses: Presented in the upper right corner is the Spearman's rank correlation comparing for all genes the two sets of log-odds of being differentially expressed. In the bottom left is given the number of genes common to the top 1,000 returned by the two methods.

Lynch et al. BMC Bioinformatics 2007 8:26   doi:10.1186/1471-2105-8-26